1 /* SPDX-License-Identifier: GPL-2.0-or-later */
3 * Linux INET6 implementation
6 * Pedro Roque <roque@di.fc.ul.pt>
12 #include <linux/ipv6.h>
13 #include <linux/hardirq.h>
14 #include <linux/jhash.h>
15 #include <linux/refcount.h>
16 #include <linux/jump_label_ratelimit.h>
17 #include <net/if_inet6.h>
19 #include <net/flow_dissector.h>
20 #include <net/inet_dscp.h>
22 #include <net/netns/hash.h>
24 struct ip_tunnel_info;
26 #define SIN6_LEN_RFC2133 24
28 #define IPV6_MAXPLEN 65535
31 * NextHeader field of IPv6 header
34 #define NEXTHDR_HOP 0 /* Hop-by-hop option header. */
35 #define NEXTHDR_IPV4 4 /* IPv4 in IPv6 */
36 #define NEXTHDR_TCP 6 /* TCP segment. */
37 #define NEXTHDR_UDP 17 /* UDP message. */
38 #define NEXTHDR_IPV6 41 /* IPv6 in IPv6 */
39 #define NEXTHDR_ROUTING 43 /* Routing header. */
40 #define NEXTHDR_FRAGMENT 44 /* Fragmentation/reassembly header. */
41 #define NEXTHDR_GRE 47 /* GRE header. */
42 #define NEXTHDR_ESP 50 /* Encapsulating security payload. */
43 #define NEXTHDR_AUTH 51 /* Authentication header. */
44 #define NEXTHDR_ICMP 58 /* ICMP for IPv6. */
45 #define NEXTHDR_NONE 59 /* No next header */
46 #define NEXTHDR_DEST 60 /* Destination options header. */
47 #define NEXTHDR_SCTP 132 /* SCTP message. */
48 #define NEXTHDR_MOBILITY 135 /* Mobility header. */
50 #define NEXTHDR_MAX 255
52 #define IPV6_DEFAULT_HOPLIMIT 64
53 #define IPV6_DEFAULT_MCASTHOPS 1
55 /* Limits on Hop-by-Hop and Destination options.
57 * Per RFC8200 there is no limit on the maximum number or lengths of options in
58 * Hop-by-Hop or Destination options other then the packet must fit in an MTU.
59 * We allow configurable limits in order to mitigate potential denial of
62 * There are three limits that may be set:
63 * - Limit the number of options in a Hop-by-Hop or Destination options
65 * - Limit the byte length of a Hop-by-Hop or Destination options extension
67 * - Disallow unknown options
69 * The limits are expressed in corresponding sysctls:
71 * ipv6.sysctl.max_dst_opts_cnt
72 * ipv6.sysctl.max_hbh_opts_cnt
73 * ipv6.sysctl.max_dst_opts_len
74 * ipv6.sysctl.max_hbh_opts_len
76 * max_*_opts_cnt is the number of TLVs that are allowed for Destination
77 * options or Hop-by-Hop options. If the number is less than zero then unknown
78 * TLVs are disallowed and the number of known options that are allowed is the
79 * absolute value. Setting the value to INT_MAX indicates no limit.
81 * max_*_opts_len is the length limit in bytes of a Destination or
82 * Hop-by-Hop options extension header. Setting the value to INT_MAX
83 * indicates no length limit.
85 * If a limit is exceeded when processing an extension header the packet is
89 /* Default limits for Hop-by-Hop and Destination options */
90 #define IP6_DEFAULT_MAX_DST_OPTS_CNT 8
91 #define IP6_DEFAULT_MAX_HBH_OPTS_CNT 8
92 #define IP6_DEFAULT_MAX_DST_OPTS_LEN INT_MAX /* No limit */
93 #define IP6_DEFAULT_MAX_HBH_OPTS_LEN INT_MAX /* No limit */
98 * type - unicast | multicast
99 * scope - local | site | global
106 #define IPV6_ADDR_ANY 0x0000U
108 #define IPV6_ADDR_UNICAST 0x0001U
109 #define IPV6_ADDR_MULTICAST 0x0002U
111 #define IPV6_ADDR_LOOPBACK 0x0010U
112 #define IPV6_ADDR_LINKLOCAL 0x0020U
113 #define IPV6_ADDR_SITELOCAL 0x0040U
115 #define IPV6_ADDR_COMPATv4 0x0080U
117 #define IPV6_ADDR_SCOPE_MASK 0x00f0U
119 #define IPV6_ADDR_MAPPED 0x1000U
124 #define IPV6_ADDR_MC_SCOPE(a) \
125 ((a)->s6_addr[1] & 0x0f) /* nonstandard */
126 #define __IPV6_ADDR_SCOPE_INVALID -1
127 #define IPV6_ADDR_SCOPE_NODELOCAL 0x01
128 #define IPV6_ADDR_SCOPE_LINKLOCAL 0x02
129 #define IPV6_ADDR_SCOPE_SITELOCAL 0x05
130 #define IPV6_ADDR_SCOPE_ORGLOCAL 0x08
131 #define IPV6_ADDR_SCOPE_GLOBAL 0x0e
136 #define IPV6_ADDR_MC_FLAG_TRANSIENT(a) \
137 ((a)->s6_addr[1] & 0x10)
138 #define IPV6_ADDR_MC_FLAG_PREFIX(a) \
139 ((a)->s6_addr[1] & 0x20)
140 #define IPV6_ADDR_MC_FLAG_RENDEZVOUS(a) \
141 ((a)->s6_addr[1] & 0x40)
144 * fragmentation header
151 __be32 identification;
154 #define IP6_MF 0x0001
155 #define IP6_OFFSET 0xFFF8
157 struct ip6_fraglist_iter {
158 struct ipv6hdr *tmp_hdr;
159 struct sk_buff *frag;
166 int ip6_fraglist_init(struct sk_buff *skb, unsigned int hlen, u8 *prevhdr,
167 u8 nexthdr, __be32 frag_id,
168 struct ip6_fraglist_iter *iter);
169 void ip6_fraglist_prepare(struct sk_buff *skb, struct ip6_fraglist_iter *iter);
171 static inline struct sk_buff *ip6_fraglist_next(struct ip6_fraglist_iter *iter)
173 struct sk_buff *skb = iter->frag;
175 iter->frag = skb->next;
176 skb_mark_not_on_list(skb);
181 struct ip6_frag_state {
194 void ip6_frag_init(struct sk_buff *skb, unsigned int hlen, unsigned int mtu,
195 unsigned short needed_tailroom, int hdr_room, u8 *prevhdr,
196 u8 nexthdr, __be32 frag_id, struct ip6_frag_state *state);
197 struct sk_buff *ip6_frag_next(struct sk_buff *skb,
198 struct ip6_frag_state *state);
200 #define IP6_REPLY_MARK(net, mark) \
201 ((net)->ipv6.sysctl.fwmark_reflect ? (mark) : 0)
203 #include <net/sock.h>
206 extern int sysctl_mld_max_msf;
207 extern int sysctl_mld_qrv;
209 #define _DEVINC(net, statname, mod, idev, field) \
211 struct inet6_dev *_idev = (idev); \
212 if (likely(_idev != NULL)) \
213 mod##SNMP_INC_STATS64((_idev)->stats.statname, (field));\
214 mod##SNMP_INC_STATS64((net)->mib.statname##_statistics, (field));\
217 /* per device counters are atomic_long_t */
218 #define _DEVINCATOMIC(net, statname, mod, idev, field) \
220 struct inet6_dev *_idev = (idev); \
221 if (likely(_idev != NULL)) \
222 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
223 mod##SNMP_INC_STATS((net)->mib.statname##_statistics, (field));\
226 /* per device and per net counters are atomic_long_t */
227 #define _DEVINC_ATOMIC_ATOMIC(net, statname, idev, field) \
229 struct inet6_dev *_idev = (idev); \
230 if (likely(_idev != NULL)) \
231 SNMP_INC_STATS_ATOMIC_LONG((_idev)->stats.statname##dev, (field)); \
232 SNMP_INC_STATS_ATOMIC_LONG((net)->mib.statname##_statistics, (field));\
235 #define _DEVADD(net, statname, mod, idev, field, val) \
237 struct inet6_dev *_idev = (idev); \
238 if (likely(_idev != NULL)) \
239 mod##SNMP_ADD_STATS((_idev)->stats.statname, (field), (val)); \
240 mod##SNMP_ADD_STATS((net)->mib.statname##_statistics, (field), (val));\
243 #define _DEVUPD(net, statname, mod, idev, field, val) \
245 struct inet6_dev *_idev = (idev); \
246 if (likely(_idev != NULL)) \
247 mod##SNMP_UPD_PO_STATS((_idev)->stats.statname, field, (val)); \
248 mod##SNMP_UPD_PO_STATS((net)->mib.statname##_statistics, field, (val));\
253 #define IP6_INC_STATS(net, idev,field) \
254 _DEVINC(net, ipv6, , idev, field)
255 #define __IP6_INC_STATS(net, idev,field) \
256 _DEVINC(net, ipv6, __, idev, field)
257 #define IP6_ADD_STATS(net, idev,field,val) \
258 _DEVADD(net, ipv6, , idev, field, val)
259 #define __IP6_ADD_STATS(net, idev,field,val) \
260 _DEVADD(net, ipv6, __, idev, field, val)
261 #define IP6_UPD_PO_STATS(net, idev,field,val) \
262 _DEVUPD(net, ipv6, , idev, field, val)
263 #define __IP6_UPD_PO_STATS(net, idev,field,val) \
264 _DEVUPD(net, ipv6, __, idev, field, val)
265 #define ICMP6_INC_STATS(net, idev, field) \
266 _DEVINCATOMIC(net, icmpv6, , idev, field)
267 #define __ICMP6_INC_STATS(net, idev, field) \
268 _DEVINCATOMIC(net, icmpv6, __, idev, field)
270 #define ICMP6MSGOUT_INC_STATS(net, idev, field) \
271 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field +256)
272 #define ICMP6MSGIN_INC_STATS(net, idev, field) \
273 _DEVINC_ATOMIC_ATOMIC(net, icmpv6msg, idev, field)
275 struct ip6_ra_chain {
276 struct ip6_ra_chain *next;
279 void (*destructor)(struct sock *);
282 extern struct ip6_ra_chain *ip6_ra_chain;
283 extern rwlock_t ip6_ra_lock;
286 This structure is prepared by protocol, when parsing
287 ancillary data and passed to IPv6.
290 struct ipv6_txoptions {
292 /* Length of this structure */
295 /* length of extension headers */
297 __u16 opt_flen; /* after fragment hdr */
298 __u16 opt_nflen; /* before fragment hdr */
300 struct ipv6_opt_hdr *hopopt;
301 struct ipv6_opt_hdr *dst0opt;
302 struct ipv6_rt_hdr *srcrt; /* Routing Header */
303 struct ipv6_opt_hdr *dst1opt;
305 /* Option buffer, as read by IPV6_PKTOPTIONS, starts here. */
308 /* flowlabel_reflect sysctl values */
309 enum flowlabel_reflect {
310 FLOWLABEL_REFLECT_ESTABLISHED = 1,
311 FLOWLABEL_REFLECT_TCP_RESET = 2,
312 FLOWLABEL_REFLECT_ICMPV6_ECHO_REPLIES = 4,
315 struct ip6_flowlabel {
316 struct ip6_flowlabel __rcu *next;
320 struct ipv6_txoptions *opt;
321 unsigned long linger;
328 unsigned long lastuse;
329 unsigned long expires;
333 #define IPV6_FLOWINFO_MASK cpu_to_be32(0x0FFFFFFF)
334 #define IPV6_FLOWLABEL_MASK cpu_to_be32(0x000FFFFF)
335 #define IPV6_FLOWLABEL_STATELESS_FLAG cpu_to_be32(0x00080000)
337 #define IPV6_TCLASS_MASK (IPV6_FLOWINFO_MASK & ~IPV6_FLOWLABEL_MASK)
338 #define IPV6_TCLASS_SHIFT 20
340 struct ipv6_fl_socklist {
341 struct ipv6_fl_socklist __rcu *next;
342 struct ip6_flowlabel *fl;
346 struct ipcm6_cookie {
347 struct sockcm_cookie sockc;
352 struct ipv6_txoptions *opt;
355 static inline void ipcm6_init(struct ipcm6_cookie *ipc6)
357 *ipc6 = (struct ipcm6_cookie) {
364 static inline void ipcm6_init_sk(struct ipcm6_cookie *ipc6,
365 const struct ipv6_pinfo *np)
367 *ipc6 = (struct ipcm6_cookie) {
369 .tclass = np->tclass,
370 .dontfrag = np->dontfrag,
374 static inline struct ipv6_txoptions *txopt_get(const struct ipv6_pinfo *np)
376 struct ipv6_txoptions *opt;
379 opt = rcu_dereference(np->opt);
381 if (!refcount_inc_not_zero(&opt->refcnt))
384 opt = rcu_pointer_handoff(opt);
390 static inline void txopt_put(struct ipv6_txoptions *opt)
392 if (opt && refcount_dec_and_test(&opt->refcnt))
396 #if IS_ENABLED(CONFIG_IPV6)
397 struct ip6_flowlabel *__fl6_sock_lookup(struct sock *sk, __be32 label);
399 extern struct static_key_false_deferred ipv6_flowlabel_exclusive;
400 static inline struct ip6_flowlabel *fl6_sock_lookup(struct sock *sk,
403 if (static_branch_unlikely(&ipv6_flowlabel_exclusive.key) &&
404 READ_ONCE(sock_net(sk)->ipv6.flowlabel_has_excl))
405 return __fl6_sock_lookup(sk, label) ? : ERR_PTR(-ENOENT);
411 struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions *opt_space,
412 struct ip6_flowlabel *fl,
413 struct ipv6_txoptions *fopt);
414 void fl6_free_socklist(struct sock *sk);
415 int ipv6_flowlabel_opt(struct sock *sk, sockptr_t optval, int optlen);
416 int ipv6_flowlabel_opt_get(struct sock *sk, struct in6_flowlabel_req *freq,
418 int ip6_flowlabel_init(void);
419 void ip6_flowlabel_cleanup(void);
420 bool ip6_autoflowlabel(struct net *net, const struct ipv6_pinfo *np);
422 static inline void fl6_sock_release(struct ip6_flowlabel *fl)
425 atomic_dec(&fl->users);
428 void icmpv6_notify(struct sk_buff *skb, u8 type, u8 code, __be32 info);
430 void icmpv6_push_pending_frames(struct sock *sk, struct flowi6 *fl6,
431 struct icmp6hdr *thdr, int len);
433 int ip6_ra_control(struct sock *sk, int sel);
435 int ipv6_parse_hopopts(struct sk_buff *skb);
437 struct ipv6_txoptions *ipv6_dup_options(struct sock *sk,
438 struct ipv6_txoptions *opt);
439 struct ipv6_txoptions *ipv6_renew_options(struct sock *sk,
440 struct ipv6_txoptions *opt,
442 struct ipv6_opt_hdr *newopt);
443 struct ipv6_txoptions *__ipv6_fixup_options(struct ipv6_txoptions *opt_space,
444 struct ipv6_txoptions *opt);
446 static inline struct ipv6_txoptions *
447 ipv6_fixup_options(struct ipv6_txoptions *opt_space, struct ipv6_txoptions *opt)
451 return __ipv6_fixup_options(opt_space, opt);
454 bool ipv6_opt_accepted(const struct sock *sk, const struct sk_buff *skb,
455 const struct inet6_skb_parm *opt);
456 struct ipv6_txoptions *ipv6_update_options(struct sock *sk,
457 struct ipv6_txoptions *opt);
459 static inline bool ipv6_accept_ra(struct inet6_dev *idev)
461 /* If forwarding is enabled, RA are not accepted unless the special
462 * hybrid mode (accept_ra=2) is enabled.
464 return idev->cnf.forwarding ? idev->cnf.accept_ra == 2 :
468 #define IPV6_FRAG_HIGH_THRESH (4 * 1024*1024) /* 4194304 */
469 #define IPV6_FRAG_LOW_THRESH (3 * 1024*1024) /* 3145728 */
470 #define IPV6_FRAG_TIMEOUT (60 * HZ) /* 60 seconds */
472 int __ipv6_addr_type(const struct in6_addr *addr);
473 static inline int ipv6_addr_type(const struct in6_addr *addr)
475 return __ipv6_addr_type(addr) & 0xffff;
478 static inline int ipv6_addr_scope(const struct in6_addr *addr)
480 return __ipv6_addr_type(addr) & IPV6_ADDR_SCOPE_MASK;
483 static inline int __ipv6_addr_src_scope(int type)
485 return (type == IPV6_ADDR_ANY) ? __IPV6_ADDR_SCOPE_INVALID : (type >> 16);
488 static inline int ipv6_addr_src_scope(const struct in6_addr *addr)
490 return __ipv6_addr_src_scope(__ipv6_addr_type(addr));
493 static inline bool __ipv6_addr_needs_scope_id(int type)
495 return type & IPV6_ADDR_LINKLOCAL ||
496 (type & IPV6_ADDR_MULTICAST &&
497 (type & (IPV6_ADDR_LOOPBACK|IPV6_ADDR_LINKLOCAL)));
500 static inline __u32 ipv6_iface_scope_id(const struct in6_addr *addr, int iface)
502 return __ipv6_addr_needs_scope_id(__ipv6_addr_type(addr)) ? iface : 0;
505 static inline int ipv6_addr_cmp(const struct in6_addr *a1, const struct in6_addr *a2)
507 return memcmp(a1, a2, sizeof(struct in6_addr));
511 ipv6_masked_addr_cmp(const struct in6_addr *a1, const struct in6_addr *m,
512 const struct in6_addr *a2)
514 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
515 const unsigned long *ul1 = (const unsigned long *)a1;
516 const unsigned long *ulm = (const unsigned long *)m;
517 const unsigned long *ul2 = (const unsigned long *)a2;
519 return !!(((ul1[0] ^ ul2[0]) & ulm[0]) |
520 ((ul1[1] ^ ul2[1]) & ulm[1]));
522 return !!(((a1->s6_addr32[0] ^ a2->s6_addr32[0]) & m->s6_addr32[0]) |
523 ((a1->s6_addr32[1] ^ a2->s6_addr32[1]) & m->s6_addr32[1]) |
524 ((a1->s6_addr32[2] ^ a2->s6_addr32[2]) & m->s6_addr32[2]) |
525 ((a1->s6_addr32[3] ^ a2->s6_addr32[3]) & m->s6_addr32[3]));
529 static inline void ipv6_addr_prefix(struct in6_addr *pfx,
530 const struct in6_addr *addr,
533 /* caller must guarantee 0 <= plen <= 128 */
537 memset(pfx->s6_addr, 0, sizeof(pfx->s6_addr));
538 memcpy(pfx->s6_addr, addr, o);
540 pfx->s6_addr[o] = addr->s6_addr[o] & (0xff00 >> b);
543 static inline void ipv6_addr_prefix_copy(struct in6_addr *addr,
544 const struct in6_addr *pfx,
547 /* caller must guarantee 0 <= plen <= 128 */
551 memcpy(addr->s6_addr, pfx, o);
553 addr->s6_addr[o] &= ~(0xff00 >> b);
554 addr->s6_addr[o] |= (pfx->s6_addr[o] & (0xff00 >> b));
558 static inline void __ipv6_addr_set_half(__be32 *addr,
559 __be32 wh, __be32 wl)
561 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
562 #if defined(__BIG_ENDIAN)
563 if (__builtin_constant_p(wh) && __builtin_constant_p(wl)) {
564 *(__force u64 *)addr = ((__force u64)(wh) << 32 | (__force u64)(wl));
567 #elif defined(__LITTLE_ENDIAN)
568 if (__builtin_constant_p(wl) && __builtin_constant_p(wh)) {
569 *(__force u64 *)addr = ((__force u64)(wl) << 32 | (__force u64)(wh));
578 static inline void ipv6_addr_set(struct in6_addr *addr,
579 __be32 w1, __be32 w2,
580 __be32 w3, __be32 w4)
582 __ipv6_addr_set_half(&addr->s6_addr32[0], w1, w2);
583 __ipv6_addr_set_half(&addr->s6_addr32[2], w3, w4);
586 static inline bool ipv6_addr_equal(const struct in6_addr *a1,
587 const struct in6_addr *a2)
589 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
590 const unsigned long *ul1 = (const unsigned long *)a1;
591 const unsigned long *ul2 = (const unsigned long *)a2;
593 return ((ul1[0] ^ ul2[0]) | (ul1[1] ^ ul2[1])) == 0UL;
595 return ((a1->s6_addr32[0] ^ a2->s6_addr32[0]) |
596 (a1->s6_addr32[1] ^ a2->s6_addr32[1]) |
597 (a1->s6_addr32[2] ^ a2->s6_addr32[2]) |
598 (a1->s6_addr32[3] ^ a2->s6_addr32[3])) == 0;
602 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
603 static inline bool __ipv6_prefix_equal64_half(const __be64 *a1,
607 if (len && ((*a1 ^ *a2) & cpu_to_be64((~0UL) << (64 - len))))
612 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
613 const struct in6_addr *addr2,
614 unsigned int prefixlen)
616 const __be64 *a1 = (const __be64 *)addr1;
617 const __be64 *a2 = (const __be64 *)addr2;
619 if (prefixlen >= 64) {
622 return __ipv6_prefix_equal64_half(a1 + 1, a2 + 1, prefixlen - 64);
624 return __ipv6_prefix_equal64_half(a1, a2, prefixlen);
627 static inline bool ipv6_prefix_equal(const struct in6_addr *addr1,
628 const struct in6_addr *addr2,
629 unsigned int prefixlen)
631 const __be32 *a1 = addr1->s6_addr32;
632 const __be32 *a2 = addr2->s6_addr32;
633 unsigned int pdw, pbi;
635 /* check complete u32 in prefix */
636 pdw = prefixlen >> 5;
637 if (pdw && memcmp(a1, a2, pdw << 2))
640 /* check incomplete u32 in prefix */
641 pbi = prefixlen & 0x1f;
642 if (pbi && ((a1[pdw] ^ a2[pdw]) & htonl((0xffffffff) << (32 - pbi))))
649 static inline bool ipv6_addr_any(const struct in6_addr *a)
651 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
652 const unsigned long *ul = (const unsigned long *)a;
654 return (ul[0] | ul[1]) == 0UL;
656 return (a->s6_addr32[0] | a->s6_addr32[1] |
657 a->s6_addr32[2] | a->s6_addr32[3]) == 0;
661 static inline u32 ipv6_addr_hash(const struct in6_addr *a)
663 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
664 const unsigned long *ul = (const unsigned long *)a;
665 unsigned long x = ul[0] ^ ul[1];
667 return (u32)(x ^ (x >> 32));
669 return (__force u32)(a->s6_addr32[0] ^ a->s6_addr32[1] ^
670 a->s6_addr32[2] ^ a->s6_addr32[3]);
674 /* more secured version of ipv6_addr_hash() */
675 static inline u32 __ipv6_addr_jhash(const struct in6_addr *a, const u32 initval)
677 u32 v = (__force u32)a->s6_addr32[0] ^ (__force u32)a->s6_addr32[1];
679 return jhash_3words(v,
680 (__force u32)a->s6_addr32[2],
681 (__force u32)a->s6_addr32[3],
685 static inline bool ipv6_addr_loopback(const struct in6_addr *a)
687 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
688 const __be64 *be = (const __be64 *)a;
690 return (be[0] | (be[1] ^ cpu_to_be64(1))) == 0UL;
692 return (a->s6_addr32[0] | a->s6_addr32[1] |
693 a->s6_addr32[2] | (a->s6_addr32[3] ^ cpu_to_be32(1))) == 0;
698 * Note that we must __force cast these to unsigned long to make sparse happy,
699 * since all of the endian-annotated types are fixed size regardless of arch.
701 static inline bool ipv6_addr_v4mapped(const struct in6_addr *a)
704 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
705 *(unsigned long *)a |
707 (__force unsigned long)(a->s6_addr32[0] | a->s6_addr32[1]) |
709 (__force unsigned long)(a->s6_addr32[2] ^
710 cpu_to_be32(0x0000ffff))) == 0UL;
713 static inline bool ipv6_addr_v4mapped_loopback(const struct in6_addr *a)
715 return ipv6_addr_v4mapped(a) && ipv4_is_loopback(a->s6_addr32[3]);
718 static inline u32 ipv6_portaddr_hash(const struct net *net,
719 const struct in6_addr *addr6,
722 unsigned int hash, mix = net_hash_mix(net);
724 if (ipv6_addr_any(addr6))
725 hash = jhash_1word(0, mix);
726 else if (ipv6_addr_v4mapped(addr6))
727 hash = jhash_1word((__force u32)addr6->s6_addr32[3], mix);
729 hash = jhash2((__force u32 *)addr6->s6_addr32, 4, mix);
735 * Check for a RFC 4843 ORCHID address
736 * (Overlay Routable Cryptographic Hash Identifiers)
738 static inline bool ipv6_addr_orchid(const struct in6_addr *a)
740 return (a->s6_addr32[0] & htonl(0xfffffff0)) == htonl(0x20010010);
743 static inline bool ipv6_addr_is_multicast(const struct in6_addr *addr)
745 return (addr->s6_addr32[0] & htonl(0xFF000000)) == htonl(0xFF000000);
748 static inline void ipv6_addr_set_v4mapped(const __be32 addr,
749 struct in6_addr *v4mapped)
751 ipv6_addr_set(v4mapped,
758 * find the first different bit between two addresses
759 * length of address must be a multiple of 32bits
761 static inline int __ipv6_addr_diff32(const void *token1, const void *token2, int addrlen)
763 const __be32 *a1 = token1, *a2 = token2;
768 for (i = 0; i < addrlen; i++) {
769 __be32 xb = a1[i] ^ a2[i];
771 return i * 32 + 31 - __fls(ntohl(xb));
775 * we should *never* get to this point since that
776 * would mean the addrs are equal
778 * However, we do get to it 8) And exacly, when
779 * addresses are equal 8)
781 * ip route add 1111::/128 via ...
782 * ip route add 1111::/64 via ...
785 * Ideally, this function should stop comparison
786 * at prefix length. It does not, but it is still OK,
787 * if returned value is greater than prefix length.
793 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
794 static inline int __ipv6_addr_diff64(const void *token1, const void *token2, int addrlen)
796 const __be64 *a1 = token1, *a2 = token2;
801 for (i = 0; i < addrlen; i++) {
802 __be64 xb = a1[i] ^ a2[i];
804 return i * 64 + 63 - __fls(be64_to_cpu(xb));
811 static inline int __ipv6_addr_diff(const void *token1, const void *token2, int addrlen)
813 #if defined(CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS) && BITS_PER_LONG == 64
814 if (__builtin_constant_p(addrlen) && !(addrlen & 7))
815 return __ipv6_addr_diff64(token1, token2, addrlen);
817 return __ipv6_addr_diff32(token1, token2, addrlen);
820 static inline int ipv6_addr_diff(const struct in6_addr *a1, const struct in6_addr *a2)
822 return __ipv6_addr_diff(a1, a2, sizeof(struct in6_addr));
825 __be32 ipv6_select_ident(struct net *net,
826 const struct in6_addr *daddr,
827 const struct in6_addr *saddr);
828 __be32 ipv6_proxy_select_ident(struct net *net, struct sk_buff *skb);
830 int ip6_dst_hoplimit(struct dst_entry *dst);
832 static inline int ip6_sk_dst_hoplimit(struct ipv6_pinfo *np, struct flowi6 *fl6,
833 struct dst_entry *dst)
837 if (ipv6_addr_is_multicast(&fl6->daddr))
838 hlimit = np->mcast_hops;
840 hlimit = np->hop_limit;
842 hlimit = ip6_dst_hoplimit(dst);
846 /* copy IPv6 saddr & daddr to flow_keys, possibly using 64bit load/store
847 * Equivalent to : flow->v6addrs.src = iph->saddr;
848 * flow->v6addrs.dst = iph->daddr;
850 static inline void iph_to_flow_copy_v6addrs(struct flow_keys *flow,
851 const struct ipv6hdr *iph)
853 BUILD_BUG_ON(offsetof(typeof(flow->addrs), v6addrs.dst) !=
854 offsetof(typeof(flow->addrs), v6addrs.src) +
855 sizeof(flow->addrs.v6addrs.src));
856 memcpy(&flow->addrs.v6addrs, &iph->saddr, sizeof(flow->addrs.v6addrs));
857 flow->control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS;
860 #if IS_ENABLED(CONFIG_IPV6)
862 static inline bool ipv6_can_nonlocal_bind(struct net *net,
863 struct inet_sock *inet)
865 return net->ipv6.sysctl.ip_nonlocal_bind ||
866 inet->freebind || inet->transparent;
869 /* Sysctl settings for net ipv6.auto_flowlabels */
870 #define IP6_AUTO_FLOW_LABEL_OFF 0
871 #define IP6_AUTO_FLOW_LABEL_OPTOUT 1
872 #define IP6_AUTO_FLOW_LABEL_OPTIN 2
873 #define IP6_AUTO_FLOW_LABEL_FORCED 3
875 #define IP6_AUTO_FLOW_LABEL_MAX IP6_AUTO_FLOW_LABEL_FORCED
877 #define IP6_DEFAULT_AUTO_FLOW_LABELS IP6_AUTO_FLOW_LABEL_OPTOUT
879 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
880 __be32 flowlabel, bool autolabel,
885 /* @flowlabel may include more than a flow label, eg, the traffic class.
886 * Here we want only the flow label value.
888 flowlabel &= IPV6_FLOWLABEL_MASK;
891 net->ipv6.sysctl.auto_flowlabels == IP6_AUTO_FLOW_LABEL_OFF ||
893 net->ipv6.sysctl.auto_flowlabels != IP6_AUTO_FLOW_LABEL_FORCED))
896 hash = skb_get_hash_flowi6(skb, fl6);
898 /* Since this is being sent on the wire obfuscate hash a bit
899 * to minimize possbility that any useful information to an
900 * attacker is leaked. Only lower 20 bits are relevant.
902 hash = rol32(hash, 16);
904 flowlabel = (__force __be32)hash & IPV6_FLOWLABEL_MASK;
906 if (net->ipv6.sysctl.flowlabel_state_ranges)
907 flowlabel |= IPV6_FLOWLABEL_STATELESS_FLAG;
912 static inline int ip6_default_np_autolabel(struct net *net)
914 switch (net->ipv6.sysctl.auto_flowlabels) {
915 case IP6_AUTO_FLOW_LABEL_OFF:
916 case IP6_AUTO_FLOW_LABEL_OPTIN:
919 case IP6_AUTO_FLOW_LABEL_OPTOUT:
920 case IP6_AUTO_FLOW_LABEL_FORCED:
925 static inline __be32 ip6_make_flowlabel(struct net *net, struct sk_buff *skb,
926 __be32 flowlabel, bool autolabel,
931 static inline int ip6_default_np_autolabel(struct net *net)
937 #if IS_ENABLED(CONFIG_IPV6)
938 static inline int ip6_multipath_hash_policy(const struct net *net)
940 return net->ipv6.sysctl.multipath_hash_policy;
942 static inline u32 ip6_multipath_hash_fields(const struct net *net)
944 return net->ipv6.sysctl.multipath_hash_fields;
947 static inline int ip6_multipath_hash_policy(const struct net *net)
951 static inline u32 ip6_multipath_hash_fields(const struct net *net)
958 * Header manipulation
960 static inline void ip6_flow_hdr(struct ipv6hdr *hdr, unsigned int tclass,
963 *(__be32 *)hdr = htonl(0x60000000 | (tclass << 20)) | flowlabel;
966 static inline __be32 ip6_flowinfo(const struct ipv6hdr *hdr)
968 return *(__be32 *)hdr & IPV6_FLOWINFO_MASK;
971 static inline __be32 ip6_flowlabel(const struct ipv6hdr *hdr)
973 return *(__be32 *)hdr & IPV6_FLOWLABEL_MASK;
976 static inline u8 ip6_tclass(__be32 flowinfo)
978 return ntohl(flowinfo & IPV6_TCLASS_MASK) >> IPV6_TCLASS_SHIFT;
981 static inline dscp_t ip6_dscp(__be32 flowinfo)
983 return inet_dsfield_to_dscp(ip6_tclass(flowinfo));
986 static inline __be32 ip6_make_flowinfo(unsigned int tclass, __be32 flowlabel)
988 return htonl(tclass << IPV6_TCLASS_SHIFT) | flowlabel;
991 static inline __be32 flowi6_get_flowlabel(const struct flowi6 *fl6)
993 return fl6->flowlabel & IPV6_FLOWLABEL_MASK;
997 * Prototypes exported by ipv6
1001 * rcv function (called from netdevice level)
1004 int ipv6_rcv(struct sk_buff *skb, struct net_device *dev,
1005 struct packet_type *pt, struct net_device *orig_dev);
1006 void ipv6_list_rcv(struct list_head *head, struct packet_type *pt,
1007 struct net_device *orig_dev);
1009 int ip6_rcv_finish(struct net *net, struct sock *sk, struct sk_buff *skb);
1012 * upper-layer output functions
1014 int ip6_xmit(const struct sock *sk, struct sk_buff *skb, struct flowi6 *fl6,
1015 __u32 mark, struct ipv6_txoptions *opt, int tclass, u32 priority);
1017 int ip6_find_1stfragopt(struct sk_buff *skb, u8 **nexthdr);
1019 int ip6_append_data(struct sock *sk,
1020 int getfrag(void *from, char *to, int offset, int len,
1021 int odd, struct sk_buff *skb),
1022 void *from, int length, int transhdrlen,
1023 struct ipcm6_cookie *ipc6, struct flowi6 *fl6,
1024 struct rt6_info *rt, unsigned int flags);
1026 int ip6_push_pending_frames(struct sock *sk);
1028 void ip6_flush_pending_frames(struct sock *sk);
1030 int ip6_send_skb(struct sk_buff *skb);
1032 struct sk_buff *__ip6_make_skb(struct sock *sk, struct sk_buff_head *queue,
1033 struct inet_cork_full *cork,
1034 struct inet6_cork *v6_cork);
1035 struct sk_buff *ip6_make_skb(struct sock *sk,
1036 int getfrag(void *from, char *to, int offset,
1037 int len, int odd, struct sk_buff *skb),
1038 void *from, int length, int transhdrlen,
1039 struct ipcm6_cookie *ipc6,
1040 struct rt6_info *rt, unsigned int flags,
1041 struct inet_cork_full *cork);
1043 static inline struct sk_buff *ip6_finish_skb(struct sock *sk)
1045 return __ip6_make_skb(sk, &sk->sk_write_queue, &inet_sk(sk)->cork,
1046 &inet6_sk(sk)->cork);
1049 int ip6_dst_lookup(struct net *net, struct sock *sk, struct dst_entry **dst,
1050 struct flowi6 *fl6);
1051 struct dst_entry *ip6_dst_lookup_flow(struct net *net, const struct sock *sk, struct flowi6 *fl6,
1052 const struct in6_addr *final_dst);
1053 struct dst_entry *ip6_sk_dst_lookup_flow(struct sock *sk, struct flowi6 *fl6,
1054 const struct in6_addr *final_dst,
1056 struct dst_entry *ip6_dst_lookup_tunnel(struct sk_buff *skb,
1057 struct net_device *dev,
1058 struct net *net, struct socket *sock,
1059 struct in6_addr *saddr,
1060 const struct ip_tunnel_info *info,
1061 u8 protocol, bool use_cache);
1062 struct dst_entry *ip6_blackhole_route(struct net *net,
1063 struct dst_entry *orig_dst);
1066 * skb processing functions
1069 int ip6_output(struct net *net, struct sock *sk, struct sk_buff *skb);
1070 int ip6_forward(struct sk_buff *skb);
1071 int ip6_input(struct sk_buff *skb);
1072 int ip6_mc_input(struct sk_buff *skb);
1073 void ip6_protocol_deliver_rcu(struct net *net, struct sk_buff *skb, int nexthdr,
1076 int __ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1077 int ip6_local_out(struct net *net, struct sock *sk, struct sk_buff *skb);
1080 * Extension header (options) processing
1083 void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1084 u8 *proto, struct in6_addr **daddr_p,
1085 struct in6_addr *saddr);
1086 void ipv6_push_frag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
1089 int ipv6_skip_exthdr(const struct sk_buff *, int start, u8 *nexthdrp,
1092 bool ipv6_ext_hdr(u8 nexthdr);
1095 IP6_FH_F_FRAG = (1 << 0),
1096 IP6_FH_F_AUTH = (1 << 1),
1097 IP6_FH_F_SKIP_RH = (1 << 2),
1100 /* find specified header and get offset to it */
1101 int ipv6_find_hdr(const struct sk_buff *skb, unsigned int *offset, int target,
1102 unsigned short *fragoff, int *fragflg);
1104 int ipv6_find_tlv(const struct sk_buff *skb, int offset, int type);
1106 struct in6_addr *fl6_update_dst(struct flowi6 *fl6,
1107 const struct ipv6_txoptions *opt,
1108 struct in6_addr *orig);
1111 * socket options (ipv6_sockglue.c)
1113 DECLARE_STATIC_KEY_FALSE(ip6_min_hopcount);
1115 int ipv6_setsockopt(struct sock *sk, int level, int optname, sockptr_t optval,
1116 unsigned int optlen);
1117 int ipv6_getsockopt(struct sock *sk, int level, int optname,
1118 char __user *optval, int __user *optlen);
1120 int __ip6_datagram_connect(struct sock *sk, struct sockaddr *addr,
1122 int ip6_datagram_connect(struct sock *sk, struct sockaddr *addr, int addr_len);
1123 int ip6_datagram_connect_v6_only(struct sock *sk, struct sockaddr *addr,
1125 int ip6_datagram_dst_update(struct sock *sk, bool fix_sk_saddr);
1126 void ip6_datagram_release_cb(struct sock *sk);
1128 int ipv6_recv_error(struct sock *sk, struct msghdr *msg, int len,
1130 int ipv6_recv_rxpmtu(struct sock *sk, struct msghdr *msg, int len,
1132 void ipv6_icmp_error(struct sock *sk, struct sk_buff *skb, int err, __be16 port,
1133 u32 info, u8 *payload);
1134 void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
1135 void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
1137 int inet6_release(struct socket *sock);
1138 int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
1139 int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
1141 int inet6_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg);
1142 int inet6_compat_ioctl(struct socket *sock, unsigned int cmd,
1145 int inet6_hash_connect(struct inet_timewait_death_row *death_row,
1147 int inet6_sendmsg(struct socket *sock, struct msghdr *msg, size_t size);
1148 int inet6_recvmsg(struct socket *sock, struct msghdr *msg, size_t size,
1154 extern const struct proto_ops inet6_stream_ops;
1155 extern const struct proto_ops inet6_dgram_ops;
1156 extern const struct proto_ops inet6_sockraw_ops;
1158 struct group_source_req;
1159 struct group_filter;
1161 int ip6_mc_source(int add, int omode, struct sock *sk,
1162 struct group_source_req *pgsr);
1163 int ip6_mc_msfilter(struct sock *sk, struct group_filter *gsf,
1164 struct sockaddr_storage *list);
1165 int ip6_mc_msfget(struct sock *sk, struct group_filter *gsf,
1166 struct sockaddr_storage __user *p);
1168 #ifdef CONFIG_PROC_FS
1169 int ac6_proc_init(struct net *net);
1170 void ac6_proc_exit(struct net *net);
1171 int raw6_proc_init(void);
1172 void raw6_proc_exit(void);
1173 int tcp6_proc_init(struct net *net);
1174 void tcp6_proc_exit(struct net *net);
1175 int udp6_proc_init(struct net *net);
1176 void udp6_proc_exit(struct net *net);
1177 int udplite6_proc_init(void);
1178 void udplite6_proc_exit(void);
1179 int ipv6_misc_proc_init(void);
1180 void ipv6_misc_proc_exit(void);
1181 int snmp6_register_dev(struct inet6_dev *idev);
1182 int snmp6_unregister_dev(struct inet6_dev *idev);
1185 static inline int ac6_proc_init(struct net *net) { return 0; }
1186 static inline void ac6_proc_exit(struct net *net) { }
1187 static inline int snmp6_register_dev(struct inet6_dev *idev) { return 0; }
1188 static inline int snmp6_unregister_dev(struct inet6_dev *idev) { return 0; }
1191 #ifdef CONFIG_SYSCTL
1192 struct ctl_table *ipv6_icmp_sysctl_init(struct net *net);
1193 struct ctl_table *ipv6_route_sysctl_init(struct net *net);
1194 int ipv6_sysctl_register(void);
1195 void ipv6_sysctl_unregister(void);
1198 int ipv6_sock_mc_join(struct sock *sk, int ifindex,
1199 const struct in6_addr *addr);
1200 int ipv6_sock_mc_join_ssm(struct sock *sk, int ifindex,
1201 const struct in6_addr *addr, unsigned int mode);
1202 int ipv6_sock_mc_drop(struct sock *sk, int ifindex,
1203 const struct in6_addr *addr);
1205 static inline int ip6_sock_set_v6only(struct sock *sk)
1207 if (inet_sk(sk)->inet_num)
1210 sk->sk_ipv6only = true;
1215 static inline void ip6_sock_set_recverr(struct sock *sk)
1218 inet6_sk(sk)->recverr = true;
1222 static inline int __ip6_sock_set_addr_preferences(struct sock *sk, int val)
1224 unsigned int pref = 0;
1225 unsigned int prefmask = ~0;
1227 /* check PUBLIC/TMP/PUBTMP_DEFAULT conflicts */
1228 switch (val & (IPV6_PREFER_SRC_PUBLIC |
1229 IPV6_PREFER_SRC_TMP |
1230 IPV6_PREFER_SRC_PUBTMP_DEFAULT)) {
1231 case IPV6_PREFER_SRC_PUBLIC:
1232 pref |= IPV6_PREFER_SRC_PUBLIC;
1233 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1234 IPV6_PREFER_SRC_TMP);
1236 case IPV6_PREFER_SRC_TMP:
1237 pref |= IPV6_PREFER_SRC_TMP;
1238 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1239 IPV6_PREFER_SRC_TMP);
1241 case IPV6_PREFER_SRC_PUBTMP_DEFAULT:
1242 prefmask &= ~(IPV6_PREFER_SRC_PUBLIC |
1243 IPV6_PREFER_SRC_TMP);
1251 /* check HOME/COA conflicts */
1252 switch (val & (IPV6_PREFER_SRC_HOME | IPV6_PREFER_SRC_COA)) {
1253 case IPV6_PREFER_SRC_HOME:
1254 prefmask &= ~IPV6_PREFER_SRC_COA;
1256 case IPV6_PREFER_SRC_COA:
1257 pref |= IPV6_PREFER_SRC_COA;
1265 /* check CGA/NONCGA conflicts */
1266 switch (val & (IPV6_PREFER_SRC_CGA|IPV6_PREFER_SRC_NONCGA)) {
1267 case IPV6_PREFER_SRC_CGA:
1268 case IPV6_PREFER_SRC_NONCGA:
1275 inet6_sk(sk)->srcprefs = (inet6_sk(sk)->srcprefs & prefmask) | pref;
1279 static inline int ip6_sock_set_addr_preferences(struct sock *sk, bool val)
1284 ret = __ip6_sock_set_addr_preferences(sk, val);
1289 static inline void ip6_sock_set_recvpktinfo(struct sock *sk)
1292 inet6_sk(sk)->rxopt.bits.rxinfo = true;
1296 #endif /* _NET_IPV6_H */